STATISTICS OF EARTHQUAKE RECURRENCE TIME IN NORTH AEGEAN TROUGH

Abstract

The determination of recurrence time of strong earthquakes of certain magnitude on a specific fault or fault segment is an important component of seismic hazard assessment. The occurrence of these earthquakes is neither periodic nor completely random but often clustered in time. This fact in connection with their limited number inhibits a deterministic approach for recurrence times calculation and thus application of stochastic processes is required. For recurrence times determination in the area of North Aegean, all the available information on strong earthquakes (historical and instrumental) with M6.0 is collected. Given that source parameters of historical events contain larger uncertainties, reassessment of their focal parameters before the application of stochastic processes is necessary, which was performed by applying the method of Bakun and Wentworth (1997). The reasses sed catalogue was divided into three data sets, according to the strong events spatial distribution and their association with distinctive fault segments. Three statistical distributions (Weibull, inverse Gaussian, lognormal) were applied and evaluated with the Anderson–Darling test and the Akaike and Bayesian Information Criteria. The Weibull distribution exhibited better performance in two out of three data sets and the Inverse Gaussian distribution in the third. With given distributions the occurrence probabilities were calculated for strong events above a certain magnitude and for certain time interval.